Academic versus reality – a tale of 2 studies (Part II)
Nico Neumann, assistant professor and fellow at the Centre for Business Analytics at Melbourne Business School, continues his look at the real impact and measures of advertising effectiveness.
Earlier we looked at the widely criticised Chicago study featured on Freakonomics, which delivered two main messages: Advertising has often been measured inaccurately and about half of CPG brands analysed seem to not achieve a positive ROI for their TV advertising investment.
Another recent study that went viral online was the work titled ‘When brands go dark’, carried out by the Australian Ehrenberg-Bass Institute (EBI). The authors collected a sample of 57 cases from alcoholic-beverage brands that switched off advertising and had declining sales in the years afterwards. This was nicely illustrated in the following graph:
What happened in adland? The graph above was extensively shared online, often earning hundreds of likes on LinkedIn posts. It was interpreted and celebrated as the ultimate evidence that brands without advertising are doomed.
Unfortunately, the devil is again in the detail and we need to be careful what conclusions we can draw. We can see that there are brands that have not advertised and that those brands – on average – had declining sales. But ask yourself: What kind of brands stop advertising for 1-9 years after having advertised before? The majority would have had strategic reasons, including brands that may have been deemed unfashionable or not showing enough success potential. So, there is some (troubled?) brand where a company has doubts and they cut advertising to reduce any losses. In other words, we don’t know which other factors – a bad product, poor distribution, competitor actions, industry headwinds etc. – have contributed to the outcome of declining sales. All we observe is that advertising has been turned off and that sales declined in the years afterwards. The EBI team highlights this in the published academic article and warns that this is not a causal analysis. This is something many are happy to overlook. Confirmation bias strikes again.
While the sample may represent a selection of many troubled brands and unique context-specific cases, we need to keep in mind again that no research is perfect. The EBI research team has done a fantastic job in finding cases where we do not require complex modelling to get some understanding of how advertising and sales may be linked. It’s a great example of the insights one can obtain by turning channels on and off.
Furthermore, some other analyses the EBI study presents seem even more thought-provoking than the first one above. For example, when looking at small, medium and big brands, we find quite different patterns illustrated in the graph below:
If we assume the graph shows a causal relationship, then this analysis suggests that big and medium brands can pause advertising for 2-3 years without losing many sales (on average). Hence, while some experts have claimed that the EBI study is evidence that Coke and other brands should not have cut advertising during the pandemic, the graph above suggests that longer ad breaks will do no serious harm to other than smaller brands.
This observation does not seem to be unique to CPG brands. When Brad Shapiro (one of the authors of the Chicago study) analysed what happened after the leading health insurance provider UnitedHealthcare ceased advertising for over 2 years in the US, he identified only a small neglectable decrease in market share.
The importance of brand size can also be seen in another analysis in the EBI study (see Chart 3). While medium and big brands that showed an upwards trajectory before stopping advertising kept growing for two years, previously growing small brands seem to lose their hard-earned sales quickly once they halt advertising.
What could explain these findings? We can compare the role of advertising to driving a car, whereby advertising can be seen as ‘accelerator’ (the gas pedal) and our spending corresponds to gasoline that we need to get going. If you are standing still (=small brand), you need lots of gas and acceleration to start moving and get speed. However, once you have momentum (=big brand), you can take your foot off the gas pedal and keep rolling at nearly the same speed for a while. Of course, you still need to accelerate every now and then to not lose all momentum. But how we drive matters a lot for fuel usage. Some drivers manage the same distance with less gas.
And here we can see that the two reviewed studies – the Chicago and EBI one – may not be too different after all, maybe even telling a similar story in different ways. Advertising typically affects sales in the short and long run, but the key question that every organisation should answer is: How much should we spend at each lifecycle of a brand to avoid wasting fuel?
No easy answers are available here and every brand is well advised to test this on their own, ideally by being brave enough to turn off advertising in an experimental setting and then see what happens.
Nico Neumann is an assistant professor and fellow at the Centre for Business Analytics at Melbourne Business School.
